Tumor to cervical spinal cord standardized uptake ratio (SUR) improves the reproducibility of F-18-FDG-PET based tumor segmentation in head and neck squamous cell carcinoma in a multicenter setting

Abstract

BACKGROUND: In quantitative FDG-PET data analysis, normalization of the standardized uptake value (SUV) with an internal image-derived standard improves its reproducibility. In this study, the cervical spinal cord is proposed as an internal standard that is within the field of view of the radiotherapy planning PET/CT-scan in head and neck cancer. The aim is to evaluate if the tumor to cervical spinal cord standardized uptake ratio (SUR) can improve the reproducibility of a model to determine the metabolic tumor volume (MTV) on FDG-PET/CT in a multicenter setting. MATERIALS AND METHODS: Ninety-five radiotherapy planning FDG-PET/CT-scans of patients with head and neck cancer were analyzed using the Bland-Altman method to evaluate differences in FDG-uptake in the cervical spinal cord and the mediastinal blood pool. Non-linear regression analysis was used to determine the optimal MTV using the gross tumor volume (GTV) as ground truth and a spatial overlap-index as statistical validation metric. Reproducibility was evaluated using the Bland-Altman method and external validation was performed in an independent dataset consisting of 62 patients. RESULTS: Bland-Altman's analyses demonstrated equivalence of FDG-uptake in the mediastinal blood pool and the cervical spinal cord. Reproducibility of the models improved when using SUR instead of SUV. These results were confirmed in the validation cohort. CONCLUSION: The use of the tumor to cervical spinal cord SUR instead of SUV improves the reproducibility of a model to determine the MTV on FDG-PET/CT in a multicenter setting. This study indicates that SUR may be preferred over SUV based approaches.